Attitudes and Cognitive Consistency* Introduction - Bertram Gawronski

In R. E. Petty, R. H. Fazio, & P. Briñol (Eds.), Attitudes: Insights from the new implicit measures. New York: Psychology Press.

4 Attitudes and Cognitive Consistency* The Role of Associative and Propositional Processes Bertram Gawronski Fritz Strack Galen V. Bodenhausen

Introduction Since the early 1950s, cognitive consistency has been a topic of continuing interest in social psychology. Notwithstanding some fundamental differences between different theories of cognitive consistency (Abelson, Aronson, McGuire, Newcomb, Rosenberg, & Tannenbaum, 1968), most of them share the assumption that cognitive inconsistency causes aversive feelings that, in turn, are assumed to have a powerful influence on judgments, decisions, and behavior. Research on cognitive dissonance (Festinger, 1957), for example, has repeatedly shown that people change their attitudes or their behavior in order to reduce the uncomfortable feeling caused by inconsistent cognitions (for an overview, see Harmon-Jones & Mills, 1999). Until recently, research investigating the impact of cognitive consistency on attitudes primarily employed explicit attitude measures. In these studies, participants were simply asked to report their attitude toward a given object. With the recent development of implicit attitude * Preparation of this chapter was supported by grants from Canada Research Chairs Program (CRC), the Social Sciences and Humanities Research Council of Canada (SSHRC), the Academic Development Fund of the University of Western Ontario (ADF), and the TransCoop Program of the Alexander von Humboldt Foundation.

Bertram Gawronski, Fritz Strack, and Galen V. Bodenhausen

Attitudes and Cognitive Consistency

measures (Fazio & Olson, 2003), however, researchers have become increasingly interested in the dynamics of cognitive consistency at the automatic level. This application of implicit attitude measures to investigate consistency phenomena was expected to improve our understanding of both implicit measures (e.g., Greenwald, Banaji, Rudman, Farnham, Nosek, & Mellott, 2002) and cognitive consistency in general (e.g., Gawronski & Strack, 2004). The main goal of the present chapter is to provide an integrative review of research on cognitive consistency employing implicit attitude measures. This review is guided by a theoretical framework proposing that implicit and explicit attitude measures tap two distinct evaluative tendencies that have their roots in qualitatively different, though interrelated, processes: associative and propositional processes (Gawronski & Bodenhausen, 2006; Strack & Deutsch, 2004). Specifically, we argue that a distinction between associative and propositional processes offers a deeper understanding of several phenomena commonly explained in terms of consistency theories, thereby providing a new perspective on how cognitive consistency influences basic attitudinal processes. In addition, we argue that an application of consistency principles to research comparing explicit and implicit attitude measures can provide deeper insights into the distinct nature of their underlying processes. For this purpose, we first outline our theoretical framework in terms of associative and propositional processes, and how cognitive consistency is related to the two kinds of mental processes. Drawing on these assumptions, we then employ the proposed distinction as an integrative framework to review research that used implicit attitude measures to study cognitive consistency.

sures-such as affective priming tasks (Fazio, Jackson, Dunton, & Williams, 1995) or the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998)-are assumed to tap evaluations that have their roots in associative processes, explicit attitude measures are assumed to tap evaluations that have their roots in propositional processes. This conceptualization resembles Eagly and Chaiken's (1993) definition of attitude as a psychological tendency to evaluate a given entity with some degree of favor or disfavor. However, the present model goes beyond this definition by arguing that such evaluative tendencies can be rooted in two different kinds of mental processes.

86

Associative and Propositional Processes The theoretical framework employed in our review is based on the Reflective-Impulsive Model (RIM; Strack & Deutsch, 2004) and its recent derivative, the Associative-Propositional Evaluation (APE) Model (Gawronski & Bodenhausen, 2006). A central notion in these models is the distinction between two qualitatively different kinds of mental processes (see also Kahneman, 2003; Lieberman, Gaunt, Gilbert, & Trope, 2002; Sloman, 1996; Smith & DeCoster, 2000). Specifically, we argue that implicitly and explicitly assessed attitudes should be understood in terms of their underlying processes. Whereas implicit attitude mea-

87

Nature of Associative and Propositional Processes

The first source of evaluative tendencies is represented by associative processes, which build the basis for evaluations reflected in implicit attitude measures. Such associative evaluations are best characterized as the automatic affective reactions resulting from the particular associations that are activated automatically upon encountering a relevant stimulus (Fazio, 1995). As such, associative evaluations require neither a high amount of cognitive capacity nor the intention to evaluate a specific object. The most important feature of associative evaluations, however, is that they are independent of the assignment of truth values (Strack & Deutsch, 2004). That is, associative evaluations can get activated irrespective of whether a person considers these evaluations as accurate or inaccurate. For example, the activation level of negative associations regarding African Americans may be high even though an individual may regard these associations as being incorrect or undesirable (Devine, 1989). The second source of evaluative tendencies resides in propositional processes, which build the basis for evaluations reflected in explicit attitude measures. Evaluations resulting from propositional processes can be characterized as evaluative judgments that have their roots in syllogistic inferences from any kind of propositional information that is considered relevant for a given judgment. In the Reflective-Impulsive Model (Strack & Deutsch, 2004), such transformations are assumed to occur in a reflective system that is superordinate to an associative store. Specifically, the reflective system is assumed to transform inputs from the associative store into propositional format (e.g., a negative affective reaction toward X is translated into the proposition "I dislike X."). The resulting propositions are then subjected to syllogistic inferences to assess their validity (Gawronski & Bodenhausen, 2006). Thus, the most



important feature that distinguishes propositional from associative processes is their dependency on truth values. Whereas the activation of associations can occur regardless of whether a person considers these associations to be true or false, propositional reasoning is generally concerned with the validation of evaluations and beliefs. Moreover, whether or not a given proposition will be explicitly endorsed depends on its subjective validity, as determined by processes of propositional reasoning.

advantaged minority members are wrong."). In this case, the resulting inconsistency between the three propositions may lead to a rejection of the negative affective reaction as a valid basis for an evaluative judgment. However, the negative affective reaction may still serve as a basis for an evaluative judgment if one of the other inconsistent propositions is rejected (Gawronski, Peters, Brochu, & Strack, 2008). The operating principles of the RIM (Strack & Deutsch, 2004) imply that propositional processes should influence associative evaluations under certain conditions. Specifically, propositional processes should influence associative evaluations when propositional reasoning leads to an affirmation of a given evaluation. However, propositional processes should leave associative evaluations unaffected when propositional reasoning leads to a negation of a given evaluation. The crucial assumption underlying this claim is that the validation process of affirming or negating a proposition implies an assignment of truth values, and thus cannot be performed associatively{Deutsch, Gawronski, & Strack, 2006). However, affirming or negating a given proposition may still activate the associative components of that proposition. Thus, affirming a propositional evaluation should directly activate its underlying associative evaluation (e.g., affirming the proposition "Old people are good drivers." activates old people and good drivers). However, negating a propositional evaluation should activate the underlying non-negated associative evaluation (e.g., negating the proposition "Old people are bad drivers." activates old people and bad drivers). Hence, negating a given proposition often leads to ironic or rebound effects on the associative level (e.g., Deutsch et al.; Gawronski, Deutsch, Mbirkou, Seibt, & Strack, 2008; Forehand & Perkins, 2005; for a review, see Wegner, 1994).2 The differential role of affirmation and negation can be illustrated with a study by Kawakami, Dovidio, Moll, Hermsen, and Russin (2000) on the reduction of automatic stereotyping. These researchers found that long-term training in the negation of social stereotypes resulted in lower levels of automatic stereotype activation. At a superficial level, this finding seems to be in contrast to the present assumptions implying that negation training should leave automatic stereotype activation unaffected. It is important to note, however, that Kawakami et al.'s negation training included two components: (a) a negation of stereotypes and (b) an affirmation of counterstereotypes. In one of their studies, for example, participants were presented with pictures of Black and White individuals and traits that were related either to the stereotype of Blacks or to the stereotype of Whites. Participants' task was to respond with a No key each time they saw a stereotype-congruent person-trait

88

Interplay Between Associative and Propositional Processes

An important aspect of the distinction between associative and propositional processes is their mutual interplay. As for the impact of associative on propositional processes, we argue that people usually base their evaluative judgments of an attitude object on their automatic affective reactions to this object (Gawronski & Bodenhausen, 2006). That is, the default mode of propositional reasoning is an affirmation of the propositional implication of an automatic affective reaction (see Gilbert, 1991). However, evaluative judgments can also be independent of automatic affective reactions when the propositional implications of these reactions are rejected as a valid basis for an evaluative judgment (Strack, 1992). Drawing on a central assumption of the APE Model (Gawronski & Bodenhausen, 2006), we argue that the primary determinant of perceived validity of a proposition-and thus of the propositional implication of an automatic affective reaction-is the consistency of this proposition with other propositions that are considered to be relevant tor the respective judgment (Kruglanski, 1989; for a discussion of alternative determinants of perceived validity, see Brifiol & Petty, 2004). In the case of evaluative judgments, such propositions may include nonevaluative propositions referring to general beliefs about the world or propositional evaluations of other attitude objects. 1 If the propositional implication of an automatic affective reaction is consistent with other relevant propositions, it may be considered valid and thus may serve as the basis for an evaluative judgment. If, however, the propositional implication of an automatic affective reaction is inconsistent with other relevant propositions, it may be considered invalid and thus may be rejected as a basis for an evaluative judgment. For example, the propositional implication of a negative affective reaction to a minority member (e.g., "I don't like this African-American person.") may be inconsistent with general beliefs about the world (e.g., "African Americans are a disadvantaged minority group.") and the propositional evaluation of another attitude object (e.g., "Negative evaluations of dis-

89

90



combination (e.g., a Black face with a stereotypically Black trait word) and to respond with a Yes key each time they saw stereotype-incongruent person-trait combination (e.g., a Black face with a stereotypically White trait word). Hence, it is not clear whether the resulting reduction in automatic stereotyping of Black people was due to the negation of the stereotype or to the affirmation of counterstereotype. Drawing on the considerations outlined above, we argue that Kawakami et al.'s findings are exclusively driven by the affirmation of the counterstereotype, rather than by the negation of the stereotype. 1his claim was recently confirmed by Gawronski, Deutsch et al. (2008), who found that only training in the affirmation of counterstereotypical information, but not training in the negation of stereotypical information, led to a reduction in automatic stereotype activation. In fact, negation of the stereotype even led to a significant increase in automatic stereotyping. This difference between affirmation versus negation is also consistent with research in other areas showing that deliberate attempts to suppress affective reactions (negation) usually leave these reactions unaffected, whereas attempts to attribute a different meaning to the response eliciting stimulus (affirmation) is indeed capable of modifying affective reactions (e.g., Butler, Egloff, Wilhelm, Smith, Erickson, & Gross, 2003; Gross, 1998).

is low. In contrast, correlations between the two kinds of measures are likely to be higher when both the motivation and the opportunity to deliberate are high (e.g., Florack, Scarabis, & Bless, 2001; Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005; Koole, Dijksterhuis, & van Knippenberg, 2001). We similarly posit that increased elaboration often reduces the correlation between explicit and implicit measures of attitudes. However, our model goes beyond the MODE model with regard to its assumptions about the underlying processes. According to the APE Model (Gawronski & Bodenhausen, 2006), cognitive elaboration primarily affects the complexity of propositional thinking by influencing how many judgment-relevant propositions are considered in addition to one's automatic affective reaction. More extensive elaboration generally implies considering a greater number of propositions about the attitude object. To the extent that any of these additional propositions is inconsistent with the automatic evaluative response, the extra elaboration will be likely to reduce the correlation between automatic affective reactions and evaluative judgments (Shiv & Nowlis, 2004). It is important to note, however, that increased cognitive elaboration does not inevitably reduce the relation between explicit and implicit attitude measures. Drawing on earlier research on directional effects of cognitive elaboration (e.g., Judd & Lusk, 1984; Petty, Brifiol, & Tormala, 2002), we argue that enhanced elaboration should reduce the relation between explicit and implicit attitude measures only if additionally considered propositions question the validity of one's automatic affective reaction as a basis for an evaluative judgment (Gawronski & Bodenhausen, 2006). However, if additionally considered propositions do not question the validity of one's automatic affective reaction, the relation between explicit and implicit attitude measures should be unaffected by cognitive elaboration. Moreover, if additionally considered propositions confirm the subjective validity of one's automatic affective reaction, the relation between explicit and implicit attitude measures should actually increase (rather than decrease) as a function of cognitive elaboration. For example, if increased cognitive elaboration identifies an additional proposition (e.g., "This African-American person behaved in a hostile manner.") that resolves the inconsistency between a propositionally transformed affective reaction (e.g., "I don't like this African-American person."), other nonevaluative propositions (e.g., "African Americans are a disadvantaged minority group."), and propositional evaluations of other attitude objects (e.g., "It is wrong to evaluate members of disadvantaged minority groups negatively."), the

Cognitive Elaboration

Cognitive elaboration has long been assumed to be of crucial importance in research on attitudes (Greenwald, 1968; Petty, Ostrom, & Brock, 1981). As with other models addressing the distinction between implicit and explicit attitude measures (e.g., Fazio & Olson, 2003; Wilson, Lindsey, & Schooler, 2000), our model implies a crucial role of cognitive elaboration for the relation between explicit and implicit attitude measures. Fazio's (1990) MODE model, for example, suggests that cognitive elaboration is a crucial determinant of the influence of automatic attitudes on behavior (see Chapter 2, this volume). Specifically, the MODE model posits that behavior is more likely to be influenced by automatic attitudes when either the motivation or the opportunity to deliberate is low. However, behavior should be less likely to be influenced by automatic attitudes when both the motivation and the opportunity to deliberate are high. Given that responses on a self-report measure simply reflect a particular kind of behavior (Fazio & Olson), the relation between explicit and implicit attitude measures is likely to be lower when either the motivation or the opportunity to deliberate

91



relation between explicit and implicit attitude measures should actually increase rather than decrease as a function of cognitive elaboration. In other words, whether the relation between explicit and implicit attitude measures increases or decreases as a function of cognitive elaboration does not depend on the amount of cognitive elaboration per se. Rather, it is a function of a consistency assessment regarding the momentarily considered set of propositions.

that have traditionally been described in terms of consistency principles. More precisely, spreading activation processes in associative networks often result in activation patterns that seem consistent from a logical perspective. For instance, if a Black person has a strong association between his or her representation of the self and the category Black, and an additional strong association between the category Black and negative, mere activation of the self should automatically activate "negative" by means of spreading activation. This spreading activation mechanism can certainly be described in propositional terms (i.e., "I am Black," "Black is bad," therefore, "I am bad."). However, this propositional description ignores that the underlying activation process is independent of whether the person considers these propositions as true or false. In other words, spreading activation processes can result in activation patterns that could be described as consistent from a propositional perspective. However, this does not necessarily imply that the process that gives raise to these activation patterns is itself propositional. We argue that the distinction between associative processes of spreading activation and propositional processes of consistency assessment is crucial when it comes to understanding the convergence and divergence between explicit and implicit attitude measures. Whereas phenomena that have their roots in associative processes of spreading activation should be more likely to emerge on implicit rather than explicit attitude measures, phenomena that have their roots in propositional processes of consistency assessment should be more likely to emerge on explicit rather than implicit attitude measures. To be sure, the fact that a given phenomenon is due to associative processes does not imply that it cannot emerge on explicit attitude measures. In fact, spreading activation should lead to the same outcome on explicit attitude measures, unless associative evaluations are rejected as a valid basis for an evaluative judgment. Conversely, the fact that a given phenomenon is due to propositional processes does not imply that it cannot emerge on implicit attitude measures. Rather, propositional processes should lead to the same outcome on implicit attitude measures when they imply an affirmation of a given evaluation, but not when they imply a negation. Importantly, even when a given process leads to corresponding effects on explicit and implicit attitude measures, spreading activation and consistency assessment should be characterized by different patterns of mediation (Gawronski & Bodenhausen, 2006). That is, if a given phenomenon has its roots in associative processes of spreading activation, this phenomenon should be characterized by a direct effect on implicit attitude measures and an indirect effect on explicit attitude measures

92

Cognitive Consistency As already outlined above, cognitive consistency plays a crucial role in the propositional process of validating evaluations and beliefs. In fact, we argue that cognitive consistency is exclusively a concern of propositional reasoning (Gawronski & Bodenhausen, 2006; Gawronski & Strack, 2004). More precisely, consistency results from a propositional process of consistency assessment that is based on the assignment of truth values and the application of syllogistic rules and logical principles.3 From a general perspective, two propositions are consistent with each other when both are regarded as true, and one does not imply the opposite of the other. In contrast, two propositions are inconsistent when both are regarded as true, and one follows from the opposite of the other (Festinger, 1957). Importantly, because (in)consistency between two propositions cannot even be defined without an assignment of truth values, inconsistency has to be resolved by means of propositional reasoning, that is, either by changing the truth value of one proposition or by finding an additional proposition that resolves the inconsistency (Gawronski & Bodenhausen, 2006). For example, if exposure to a minority member automatically activates negative associations, people may either reject the propositional implication of these associations because of its inconsistency with other accepted propositions (Gawronski, Peters et al., 2008), or they may find an additional proposition that resolves the inconsistency (e.g., "This African-American person was unfriendly."). Whereas the former process refers to what we described as negation of the propositional implications of an automatic affective reaction (Deutsch et al., 2006), the latter process has been described as rationalization (Festinger) or justification (Crandall & Eshleman, 2003). Notwithstanding the propositional nature of cognitive consistency, it is important to note that associative processes can produce outcomes

I 1

93

94



that is mediated by the effect on implicit attitude measures. In contrast, if a given phenomenon has its roots in propositional processes of consistency assessment, this phenomenon should be characterized by a direct effect on explicit attitude measures and an indirect effect on implicit attitude measures that is mediated by the effect on explicit attitude measures.

women associated the category women with a positive (negative) evaluation, and the stronger they associated themselves with the category women, the more positive (negative) was their automatic self-evaluation. Interestingly, such patterns of balanced identities were generally obtained for implicit measures, whereas identities often showed imbalanced patterns on explicit measures. Drawing on the distinction between associative and propositional processes, these findings can be explained in terms of spreading activation in associative memory. Specifically, the activation of a particular